1. Technical Field
The present invention relates to an image processing device, an imaging device, a computer-readable storage medium, and an image processing method, and in particular, relates to an imaging device, an image processing device, a computer-readable storage medium, and an image processing method that carry out image processing on images obtained by capturing images continuously.
2. Related Art
Conventionally, a technique, that generates a captured image having low noise by carrying out imaging plural times continuously and making an image, that is obtained by aligning and combining the plural images that are obtained by the imaging, be the final captured image, is used as a technique of capturing a highly-detailed image by an imaging device. At the time of combining the images in this technique, motion vectors (amounts of movement) between the respective images that are the objects of combining are determined by a gyro sensor or image processing, and the images are combined after carrying out aligning by parallel-translating or rotating the plural images on the basis of the motion vectors.
However, in cases such as when a person carries out imaging by holding an imaging device in his/her hands or the like, there are cases in which, due to camera shaking, the image angles between respective images differ relatively greatly even in images that are captured continuously. Further, by transforming images that are objects of combining, pixel regions in which the images are combined together and pixel regions in which the images are not combined together arise at the time of combining the images. Therefore, in a case of combining plural images, the number of times of combining differs per pixel region in the composite image. On the other hand, at the time of combining plural images, there are also cases in which control is carried out so as to not combine pixels corresponding to regions where subject blurring has occurred or regions where occlusion has occurred or the like. In such cases as well, the number of times of combining differs per pixel region in the composite image.
When the number of times of combining differs per pixel region in this way, dispersion arises per pixel region in the noise reduction amounts in the composite image. Namely, even if noise is sufficiently reduced at a pixel region where the number of times of combining is large, there is the possibility that noise will remain at a pixel region where the number of times of combining is small.
In order to overcome this, Japanese Patent Application Laid-Open (JP-A) No. 2009-194700 discloses an imaging device that carries out noise reduction processing, that corresponds to the number of time of combining per pixel region, on a composite image. This imaging device has: number of times of addition computing unit that, after aligning plural images that have been captured continuously, adds the images and generates a composite image, and computes the number of times of addition per pixel of the composite image; and noise reducing unit that reduces noise in accordance with the number of times of addition per pixel of the composite image. Owing to this structure, it is possible to generated a highly-detailed image.
In the technique disclosed in JP-A No. 2009-194700, the noise reduction processing is carried out in accordance with the number of times of combining per pixel region of the composite image. However, the noise reduction processing function that a general imaging device is equipped with is a function that carries out noise reduction processing with the entire image being the object thereof. Namely, in order to reduce noise per pixel region of a composite image, it is necessary to separately provide a function that designates a pixel region and carries out noise reduction processing thereon.